The Agent Skills Directory

[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it reads and processes external source code files during cleanup operations.
Ingestion points: Local files identified via git diff or user specification in SKILL.md.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to isolate file content from the agent's execution context.
Capability inventory: The skill utilizes file modification capabilities for code cleanup and command execution for running test suites.
Sanitization: Absent; source code content is processed directly without specific validation or filtering.
[COMMAND_EXECUTION]: The skill uses shell commands for project introspection and verification.
Evidence: The workflow in SKILL.md involves executing git diff to identify modified files and running user-defined test suites to verify the safety of logic refactoring in Tier 2.

boy-scout